Human society is characterized by generalized exchange (i.e., unilateral resource giving among n-person). Recent theoretical studies have tried to discover how generalized exchange can emerge and be maintained (e.g., Nowak and Sigmund, 1998; Ohtsuki and Iwasa, 2005; Panchanathan and Boyd, 2003; Takahashi and Mashima, 2003). Those studies contain the same idea – the existence of discriminate altruism is the key that makes generalized exchange possible. However, these studies differ in their answer to the question of “what type of discriminative strategy is the solution”. Furthermore, there are few empirical studies that examine people’s strategies in generalized exchange. The purpose of the current study is to discover what type of person people regard as “good” (i.e. worth giving to), and what type of person people regard as “bad” (i.e. not worth giving to) in generalized exchange.
We conducted a vignette study to reveal strategies that people actually use. The respondents were asked to read several scenarios of generalized exchange, and then answer several questions about the target person who was described in each scenario. Each scenario described what the target person (a potential donor) did to two potential recipients. We manipulated two factors – first-order information and the second-order information. First-order information is the behavior of the target (gave to a “good” recipient, gave to a “bad” recipient, or didn’t give to anyone). Second-order information is the reputation of potential recipients (“good” or “bad”). After reading the scenario, respondents gave their impressions of the target person and their willingness to give resources to that person.
Results showed that respondents evaluated 1) the target who gave to a good recipient as “good”, 2) the target who gave to a bad recipient as “bad”, and 3) the target who did not give to a good recipient as “bad”. Willingness to give to the target showed a qualitatively similar pattern. Willingness to give to the target was highest toward the target who gave to a good recipient, midway toward the target who gave to a bad recipient, and lowest toward the target who did not give to a good recipient. These results are consistent with Takahashi and Mashima’s (2003) conclusion, that the key point that makes generalized exchange possible is to exclude or punish not only free-riders but also indiscriminate givers who help free-riders.